The crucial role the Crumbs and Par polarity complexes play in tight junction integrity has long been established, however very few studies have investigated the role of the Scribble polarity module. Here, we use MCF10A cells, which fail to form tight junctions and express very little endogenous Crumbs3, to show that inducing expression of the polarity protein Scribble is sufficient to promote tight junction formation. We show this occurs through an epithelial-to-mesenchymal (EMT) pathway that involves Scribble suppressing ERK phosphorylation, leading to downregulation of the EMT inducer ZEB. Inhibition of ZEB relieves the repression on Crumbs3, resulting in increased expression of this crucial tight junction regulator. The combined effect of this Scribble-mediated pathway is the upregulation of a number of junctional proteins and the formation of functional tight junctions. These data suggests a novel role for Scribble in positively regulating tight junction assembly through transcriptional regulation of an EMT signaling program.
The formation and maintenance of apically located tight junctions is key for the establishment of correct tissue architecture and epithelial cell polarity. Major functions attributed to tight junctions include the regulation of paracellular permeability and a ‘fence’ or physical barrier function to control the diffusion of lipids and proteins within the plasma membrane thus contributing to the organization of the apical and basal domains (Ebnet, 2008; Shin et al., 2006).
There are three evolutionary conserved polarity complexes; the Scribble, Par and Crumbs complexes that act as regulators of apical-basal polarity. Members of each of these complexes have been reported to play important roles in establishment and maintenance of tight junctions (Chen and Macara, 2005; Fogg et al., 2005; Hurd et al., 2003; Ivanov et al., 2010; Joberty et al., 2000; Lemmers et al., 2004; Michel et al., 2005; Qin et al., 2005; Shin et al., 2005; Straight et al., 2004; Stucke et al., 2007; Suzuki et al., 2001; Suzuki et al., 2009). Most of these studies have focused on the Par and Crumbs complexes, which are intimately associated with tight junctions however, more recently an important role for the Scribble complex has also emerged. Scribble knockdown in MDCK cells results in a delay in tight junction assembly and impaired recruitment of E-cadherin to the membrane (Qin et al., 2005), and loss of Scribble in human colon cells has been shown to impair tight junction assembly (Ivanov et al., 2010).
Members of the Scribble complex, Scribble, Discs Large (Dlg) and Lethal Giant Larvae (Lgl), were all identified in Drosophila melanogaster as neoplastic tumor suppressors, with loss of function disrupting apical basal polarity and junctional integrity, and inducing inappropriate proliferation and tissue overgrowth (Bilder and Perrimon, 2000; Mechler et al., 1985; Woods and Bryant, 1991). Since then, Scribble has been implicated in numerous cellular processes including proliferation, differentiation, apoptosis, stem-cell maintenance, migration and vesicle trafficking (Elsum et al., 2012; Humbert et al., 2008). Human Scribble is a functional homologue of Drosophila Scribble (Dow et al., 2003) and is a target for ubiquitin-mediated degradation by the human papilloma virus (HPV) E6 proteins and E6AP protein ligase (Nakagawa and Huibregtse, 2000). HPV is frequently associated with cervical carcinomas, which often show decreased Scribble expression (Nakagawa et al., 2004). Other oncogenic viruses such as human T-cell leukemia virus type 1 (HTLV-1) Tax, a causative agent for adult T-cell leukemia, are known to target Scribble (Elsum et al., 2012; Javier and Rice, 2011; Okajima et al., 2008). Additionally, mislocalization or deregulated Scribble expression has been reported in a variety of epithelial cancers (Gardiol et al., 2006; Kamei et al., 2007; Navarro et al., 2005; Ouyang et al., 2010; Pearson et al., 2011; Vaira et al., 2011; Zhan et al., 2008). The mechanism by which Scribble influences tumorigenesis is unclear, yet may lie in the interactions with oncogenic signaling cascades such as Ras, Wnt and GPCR signaling (Humbert et al., 2008).
Tight junctions are recognized to have important roles in acting as signaling platforms in a variety of cellular processes (Balda and Matter, 2008; Balda and Matter, 2009; Farkas et al., 2012; González-Mariscal et al., 2008; Steed et al., 2010; Yang et al., 2012). In turn, components of numerous intracellular signaling pathways including G-proteins, PLCγ and PKC have been implicated in the regulation and maintenance of tight junction integrity (Nunbhakdi-Craig et al., 2002; Nusrat et al., 1995; Stuart and Nigam, 1995; Terry et al., 2011; Walsh et al., 2001; Ward et al., 2002). Of note, tight junction regulation is among the plethora of cellular processes the Ras MAPK (mitogen activated protein kinase)-ERK (extracellular-signal-regulated kinase) pathway is involved in. The MAPK-ERK cascade involves a series of phosphorylation events to regulate machinery controlling physiological processes such as proliferation, growth, apoptosis and migration (Dhillon et al., 2007; Shaul and Seger, 2007). The role MAPK-ERK signaling plays in tight junction integrity appears complex and context dependent. In intestinal cells, MAPK-ERK activation results in increased expression of crucial tight junction proteins (Yang et al., 2005). In Caco2 cells, MAPK-ERK signaling regulates oxidative-stress-induced disruption of tight junctions in a manner that is dependent on their differentiation state (Aggarwal et al., 2011; Basuroy et al., 2006). In Ras transformed MDCK cells, inhibition of the MAPK-ERK pathway restores tight junctions (Chen et al., 2000) and expression of a constitutively active Raf-1 in a rat epithelial cell line promotes downregulation of the tight junction proteins Occludin and Claudin1 (Li and Mrsny, 2000). Although the precise mechanisms remain unclear, these and numerous other examples illustrate that MAPK-ERK signaling can influence junction dynamics.
Epithelial-to-mesenchymal transition (EMT) signaling is among the numerous pathways mediated by MAPK-ERK signaling. Recently ERK2 was shown to regulate the expression of ZEB through Fos related antigen-1 (Fra1) (Shin et al., 2010). The zinc finger transcription factors ZEB1 and ZEB2 (also known as δEF1 and SIP1 respectively) are classic examples of EMT-induced transcriptional regulators that are activated during tumor cell dissemination and invasion (Schmalhofer et al., 2009; Spaderna et al., 2008). They have been shown to suppress tight junction proteins including Occludin, Claudins, Tricellulin, JAM and ZO3 (Aigner et al., 2007; Comijn et al., 2001; Peinado et al., 2007; Vandewalle et al., 2005), as well as several polarity proteins, such as Crumbs3, PATJ and Lgl2, through direct binding to E-boxes in the promoter regions (Aigner et al., 2007; Spaderna et al., 2008). Fra1, a member of the Fos family and a key component of the AP-1 transcriptional complex, is reported to be involved in tumor progression and invasion in a variety of different human tumor cell lines (Young and Colburn, 2006). Furthermore, activation of the MAPK-ERK pathway is known to increase expression of Fra1 with ERK-mediated phosphorylation stabilizing Fra1 to promote efficient activity (Hoffmann et al., 2005; Treinies et al., 1999; Young et al., 2002).
We have previously shown that Scribble mediates MAPK-ERK signaling both in vitro and in vivo (Dow et al., 2008; Pearson et al., 2011). Here we demonstrate the existence of a pathway that encompasses both polarity and signaling molecules to regulate tight junction formation through regulation of MAPK-ERK activity. Using MCF10A cells, a mammary epithelial cell line that despite forming desmosomes and adherens junctions, lack tight junctions under standard tissue culture conditions (Underwood et al., 2006), we show that increased expression of Scribble promotes the formation of functional tight junctions. We provide evidence for a pathway in which Scribble regulates MAPK-ERK signaling which in turn mediates the expression of the EMT inducer ZEB via Fra1. Inhibition of Fra1 leads to a reduction in ZEB levels and consequently an increase in the polarity protein Crumbs3, an essential component of tight junctions. The data presented here show a novel role for Scribble in positively regulating tight junction assembly through modulation of an EMT signaling and transcriptional program.
Scribble promotes tight junction formation
To study the effect of aberrant Scribble expression in MCF10A cells, stable lines were generated by retroviral transduction in which human Scribble was knocked down or overexpressed. Extensive microarray analysis was performed to investigate how Scribble expression affects the global transcriptional programme (I.A.E., unpublished data). We found that cells ectopically expressing Scribble (ScribOE) at levels approximately ten-fold greater than vector controls (Fig. 1B) had a strong cell adhesion signature. DAVID bioinformatics analysis (http://david.abcc.ncifcrf.gov/) showed enrichment in the KEGG ‘cell adhesion molecules’ pathway with a P-value of 7.9×10−2. Examination of the gene list (supplementary material Table S1) indicated many proteins associated with desmosomes and tight junctions, including Occludin and Claudin1, were upregulated by Scribble overexpression (Fig. 1A). Western blot analysis confirmed this upregulation occurred at a protein level (Fig. 1B).
ZO1, a scaffolding protein associated with tight junctions, can be used as a marker of junction integrity. We found localization of ZO1 to be dramatically altered by Scribble overexpression (Fig. 1C) yet total expression levels remained unchanged (Fig. 1B). Consistent with previous reports (Fogg et al., 2005), vector control MCF10A cells displayed a diffuse pattern of ZO1, with punctate staining around the membrane. In contrast, ScribOE cells displayed smooth, continuous ZO1 staining, typical of intact tight junctions, and quantification of ZO1 staining showed a significant 3.3-fold increase in unbroken continuous staining in ScribOE cells compared to control cells (Fig. 1C). Immunostaining for other tight junction markers, Occludin and Claudin1, also showed an increase in continuous staining in ScribOE cells, supporting the ZO1 staining (Fig. 1C). It is interesting to note that despite being a stable population with all cells expressing Scribble, only a subset of the ScribOE cells localize ZO1, Occludin or Claudin1 to junctions. The patchy nature of the rescued tight junction protein localization is puzzling, however we observed this in stable cell lines expressing Scribble as well as following use of chemical inhibitors and siRNA (see below). This suggests that additional extrinsic factors such as high cell confluency may also enhance ZO1 restoration in ScribOE cells at a population level.
To functionally characterize Scribble-mediated promotion of adhesion, we carried out a range of assays including transmission electron microscopy (TEM), transepithelial resistance (TER) and size-selective assessment of tight junction paracellular permeability using fluorescently labelled dextrans (PPFD). Analysis of the microstructure of the epithelial sheet in MCF10A vector and ScribOE cells using TEM revealed an increase in adhesion and cell junctions in ScribOE cells (supplementary material Fig. S1A). Additionally, we carried out TER measurements to determine the ion permeability of assayed junctions. ScribOE cells were found to have a modest yet significant increase in TER when grown for 13 days on transwell filters (Fig. 1D). As vector control and wild-type MCF10A cells lack tight junctions, readings were similar to the blank wells. However, Scribble overexpressing cells recorded values ∼1.5-fold higher than vector cells. MDCK cells, which form complete, intact tight junctions, were used as a positive control and produced TER readings close to 500 ohms.cm2 (data not shown). Tight junctions regulate diffusion of ions and molecules both by charge and size and so, to effectively measure tight junction functionality, both must be measured. TER assays determined the diffusion of ions and to assay size diffusion we performed size-selective assessment of tight junction paracellular permeability using fluorescently labelled dextrans (PPFD) (Matter and Balda, 2003). Vector control and ScribOE cells were grown for 13 days on transwell filters prior to addition of fluorescently labeled 70 kDa Rhodamine and 4 kDa FitC dextran molecules (Fig. 1E). As the experimental values in a sample group varied despite the trends between sample groups remaining consistent, the data has been shown here as the change in fluorescent diffusion in the ScribOE cells relative to the vector cells. A significant difference was seen in the diffusion of the larger 70 kDa molecule yet not the smaller 4 kDa suggesting that the rudimentary tight junctions formed in cells overexpressing Scribble are more successful in blocking larger molecules. As MDCK cells have well-formed tight junctions, they were used as a positive control and showed values close to zero across both sizes of dextran.
Inhibition of MAPK-ERK signaling promotes tight junction assembly
We have previously shown that Scribble overexpression suppresses MAPK activity in MCF10A cells in the context of oncogenic Ras (RasV12) (Dow et al., 2008). We carried out biochemical analysis on ScribOE cells to further look at how altered Scribble expression alone can influence MAPK signaling. We found that in resting cells, Scribble overexpression suppresses phosphorylation at multiple tiers of the MAPK-ERK pathway including MEK and ERK [Fig. 2A(i)]. Upon EGF stimulation, the phosphorylation response of MEK was also muted [Fig. 2A(ii)]. Given this, we were interested in looking at the role of MAPK signaling in tight junction formation in MCF10A cells. By using specific inhibitors against MEK1 (PD98059), PI3K (LY294002) and JNK (SP600125) (supplementary material Fig. S1B) we were able to assess the individual contributions of each of these pathways to tight junction formation. Of the tested pathways, only inhibition of the MEK arm was sufficient to promote tight junction formation (Fig. 2B).
In agreement with these results, siRNA-mediated knockdown of ERK1 and ERK2 was found to promote the formation of tight junctions (Fig. 2C). Quantification of staining revealed a significant 3.8-fold increase in continuous ZO1 staining in siERK1/2 cells compared to the siControl [Fig. 2C(ii)]. No detectable differences were seen in ZO1 staining between cells individually transfected with siERK1 and siERK2 or combined transfections (supplementary material Fig. S1D). Further confirmation of tight junction formation was seen by staining for Occludin and Claudin1 in cells where both ERK1 and ERK2 had been depleted (Fig. 4B).
Inhibition of Fra1 and ZEB1/2 promotes tight junction assembly
ERK2 has been reported to regulate expression of the EMT inducer ZEB, through Fra1 (Shin et al., 2010). We investigated whether Scribble expression influences ZEB expression through such a pathway.
Overexpression of Scribble resulted in a decrease in Fra1 at both the protein and mRNA level and a decrease in Fra1 was also seen upon knockdown of ERK1/2 (Fig. 3A). As ZEB has been shown to lie downstream of Fra1 (Shin et al., 2010), we looked at whether changes in upstream components such as Scribble or ERK could alter ZEB expression and consequently tight junction formation. Forced Scribble expression, siERK1/2 or siFra1 all resulted in downregulation of ZEB1 and ZEB2 (Fig. 3B). To test the functional consequences of these findings, MCF10A cells were depleted of Fra1 or ZEB1/2 using siRNAs (supplementary material Fig. S2) and stained for ZO1, Occludin and Claudin1 (Fig. 3C,D). In all cases knockdown modulated tight junction formation and quantification of continuous ZO1 staining showed an approximate 1.5- and 3-fold increase for Fra1 and ZEB respectively compared to the controls.
Given that EMT inducers are known to influence tight junction integrity (Aigner et al., 2007; Ikenouchi et al., 2003; Vandewalle et al., 2005) we were interested in investigating whether enforced Scribble expression resulted in altered expression of EMT inducers other than ZEB. We found that similarly to the changes in ZEB expression, Snail and Slug mRNA levels were reduced in ScribOE cells. However, unlike ZEB, depletion of either Snail or Slug was insufficient to promote the formation of tight junctions (supplementary material Fig. S3). Together these data suggest a specific, functional role for the ZEB proteins in tight junction assembly in MCF10A cells.
Crumbs3 is required for Scribble and MAPK-dependent tight junction formation
Exogenous expression of Crumbs3 in MCF10A cells is sufficient to induce tight junction structures (Fogg et al., 2005). As several studies have identified Crumbs3 as a ZEB target (Aigner et al., 2007; Spaderna et al., 2008), we investigated the possibility of a ZEB dependent involvement for Crumbs3 in Scribble-mediated tight junction formation. We first confirmed that ectopic Crumbs3 expression induced tight junction structures. Stable MCF10A cell lines were generated using retroviral transduction to express Crumbs3 at levels ∼15–20 times greater than vector controls (supplementary material Fig. S2). Consistent with previous reports (Fogg et al., 2005), cells overexpressing Crumbs3 (Crumbs3OE) had smooth continuous ZO1 and Occludin staining compared to vector controls, indicative of the formation of tight junctions (supplementary material Fig. S4A). Additionally, TEM analysis demonstrated the presence of tight junctions in Crumb3OE cells yet only desmosomes were seen in vector control cells (supplementary material Fig. S4B). TER measurements showed an increase of more than 4-fold in Crumb3OE cells compared to control cells (supplementary material Fig. S4C). Size diffusion was assayed using PPFD and the percentage of molecules that had diffused through the tight junctions in the Crumb3OE cells were compared to the vector control cells. A larger and more significant difference was seen with the diffusion of the larger 70 kDa dextran molecule rather than the smaller 4 kDa molecule (supplementary material Fig. S4D). This is a similar trend as to what was observed in the ScribOE cells (Fig. 1E) and once more suggests that the Crumb3OE cells partially block larger molecules and the smaller dextran molecules to a lesser extent. We also looked at an earlier time point to address if there was a difference in the kinetics of tight junction formation in the ScribOE and Crumb3OE cells. We performed PPFD 3 days after plating onto transwell filters. Again there was a significant difference in the diffusion of the large 70 kDa dextran molecule but not the smaller 4 kDa molecule in both the ScribOE and Crumb3OE cells (supplementary material Fig. S5). There was a greater difference in the diffusion of both the small and large dextrans in the Crumb3OE cells than in the ScribOE cells suggesting that the former have more intrinsically intact tight junctions.
To determine if Scribble expression or MAPK-ERK signaling influences Crumbs3 transcription via a Fra1/ZEB pathway we measured Crumbs3 mRNA in ScribOE cells, cells treated with siERK1/2, siFra1 or siZEB1/2. All conditions resulted in modest yet significant increases in Crumbs3 (Fig. 4A). Despite extensive efforts, we were unfortunately unable to measure the protein levels and localization of Crumbs3 in ScribOE cells, cells treated with siERK1/2, siFra1 or siZEB1/2 as the available commercial antibodies did not reliably detect endogenous Crumbs3. To test the requirement of Crumbs3 in the tight junction phenotype observed in ScribOE cells, siERK1/2, siFra1 and siZEB1/2 cells, we made use of siRNAs targeted against Crumbs3 (supplementary material Fig. S2). Co-transfection of siCrumbs3 in ScribOE, siERK1/2, siFra1 or siZEB1/2 cells abrogated the formation of tight junctions (Fig. 4B) indicative of an essential role for Crumbs3 in tight junction assembly in MCF10A cells. Of note, Scribble expression or localization was not altered in the presence of siCrumbs3 (data not shown). Taken together, our results support a model in which Scribble acts to regulate tight junction formation in MCF10A cells through a pathway involving MAPK-ERK signaling, expression of ZEB, Fra1 and Crumbs3 (Fig. 5).
Active EMT programs are known to deregulate the expression of many essential polarity and junction proteins, resulting in polarity loss and junction dissolution (Godde et al., 2010; Moreno-Bueno et al., 2008; Peinado et al., 2007). In this study we build on this notion and have identified the polarity protein Scribble as a novel regulator of an EMT signaling and transcriptional pathway involved in the establishment of tight junctions. We have shown that Scribble-mediated suppression of MAPK-ERK signaling results in downregulation of the ZEB proteins through changes to the expression of the gene product Fra1. Consequently, the polarity protein Crumbs3 is transcriptionally upregulated leading to the formation of functional tight junctions. The data presented here also indicates that Scribble is likely to act on additional pathways that regulate tight junction integrity to that described above as overexpression of Scribble also results in an increase in Occludin and Claudin1 expression (Fig. 1A), this effect is not seen upon Crumbs3 overxpression (Fogg et al., 2005). It is possible that the upregulation of Occludin and Claudin1 and the upregulation of Crumbs3 are processes that lie parallel and separately contribute to tight junction formation.
As mentioned previously, ERK2 has been shown to regulate the expression of ZEB through Fra1 (Shin et al., 2010). Shin and colleagues demonstrated that ERK2, but not ERK1, specifically mediates upregulation of ZEB and induction of an EMT phenotype. Although we see no difference in efficacy of tight junction formation with siERK1 vs siERK2 (supplementary material Fig. S1D), we see a greater increase in Crumbs3 expression and a greater decrease in Fra1 and Zeb1 expression with ERK2 depletion than ERK1 (data not shown). These results are therefore consistent with Shin et al. (Shin et al., 2010) and suggest that ERK2 is playing a greater role than ERK1 in controlling the EMT phenotype, however this does not correlate to changes in tight junction integrity and suggests that other pathways than EMT signaling contribute to aspects of tight junction integrity.
What other Fra1 and Zeb targets may be involved in tight junction regulation? Fra1 is part of the AP-1 transcriptional complex that interacts with several proteins including retinoblastoma, SMAD3, SMAD4 and the p65 subunit of NFκB (Young and Colburn, 2006). In addition to MAPK signaling, regulation of ZEB expression has been linked to hormones, TGFβ and also of note, NFκB signaling (Chua et al., 2007; Dillner and Sanders, 2004; Dohadwala et al., 2006; Peinado et al., 2007; Shirakihara et al., 2007). As overexpression of the NFκB subunit p65 in MCF10A cells results in increased expression of ZEB1 and ZEB2 (Chua et al., 2007), these data suggest that ZEB and Fra1 could mediate part of their effects through regulation of NFκB pathway. Nevertheless, we found that depletion of various components of the NFκB pathway in MCF10A cells had no discernible effect on tight junction formation (supplementary material Fig. S6). This data strengthens our hypothesis for a specific role of the ERK→Fra1→ZEB→Crumbs3 pathway in tight junction assembly (Fig. 5).
Our laboratory has previously shown that Scribble suppresses Ras dependent transformation through inhibition of MAPK activity (Dow et al., 2008). It is likely that in ScribOE MCF10A cells, Scribble-mediated suppression of MAPK-ERK signaling prevents stabilization of Fra1 and results in the decreased ZEB expression reported here. This raises the question of how Scribble is suppressing MAPK-ERK signaling. Recently a direct protein interaction between Scribble and ERK was reported (Nagasaka et al., 2010b). In this study the authors suggest that ERK activation enhances the interaction between Scribble and ERK, the formation of such a Scribble–ERK complex then acts to inhibit MAPK activity. Varying phosphorylation sites within Scribble have been proposed to alter its ability to bind ERK and the cellular localization (Nagasaka et al., 2010a; Yoshihara et al., 2011). Phosphorylation events within Scribble may therefore alter its function thus allowing it to function as a modulator of signaling and/or a scaffold. Scribble may act to differentially regulate several cellular processes in a manner that is dependent on the level of activity of particular signaling pathway. When Scribble is ectopically expressed, as reported here, the system may become saturated so that Scribble binds ERK, forming a complex to alter downstream signaling events including inhibition of EMT programs while additionally functioning at other cellular levels to promote the formation of tight junctions. Such functions may include acting as a dynamic scaffolding molecule by interacting with known binding partners such as ZO2, βPix and β-catenin (Audebert et al., 2004; Métais et al., 2005; Sun et al., 2009). Recently Scribble was shown to recruit the phosphatase PHLPP1 to the cell membrane to negatively regulate AKT signaling (Li et al., 2011). We and others have previously shown that Scribble regulates the phosphorylation status of several components of MAPK signaling (Dow et al., 2008; Elsum and Humbert, 2013; Nagasaka et al., 2010b). Precisely how Scribble interacts with phosphatases to influence signaling and junction dynamics is a question that needs to be further addressed.
A key downstream consequence of the Scribble-mediated suppression of MAPK signaling demonstrated here is the increased expression of the apical polarity gene Crumbs3. Several studies performed in cell lines with either microdeletions or low levels of endogenous Crumbs3 have shown that reconstituting with ectopic Crumbs3 is sufficient to promote the formation of tight junctions (Fogg et al., 2005; Karp et al., 2008; Rothenberg et al., 2010). Furthermore, downregulation of Crumbs3 by EMT inducers such as Snail results in loss of tight junctions (Whiteman et al., 2008). The functional role Crumbs3 plays in tight junctions is unclear yet it is likely to be involved in recruiting or stabilizing other protein complexes critical for tight junction formation.
In summary, we have found that expression of Scribble in MCF10A cells results in the formation of tight junctions. We have provided evidence that Scribble acts via a MAPK-ERK→Fra1→ZEB→Crumbs3 axis to regulate tight junctions. Although this study has focused on tight junctions as a physiological read out, these findings have far broader implications regarding how Scribble and other polarity proteins may feed into the MAPK pathway and influence tumor progression through modulation of other processes where EMT has been implicated such as invasion and metastasis. Notably, the present study is among the first to describe how one polarity protein can transcriptionally regulate another. This is important in understanding how interactions and feedback within the polarity network occur. This study and some of the examples outlined above, illustrate that a finely tuned balance between different polarity proteins is required for the correct establishment and maintenance of tight junctions.
Materials and Methods
Immunostaining and western blotting were performed with the following antibodies: rabbit anti-pAKT (ser473), anti-total AKT, anti-total ERK1/2, anti-total JNK, anti-pMEK1/2 (41G9) and mouse anti-pERK1/2 (Thr202/Tyr204), anti-total MEK1/2 all purchased from Cell Signaling, Danvers, MA. Rabbit anti-Claudin1 (51-9000 Zymed, San Francisco, CA), rabbit anti-occludin (71-1500 Zymed), rabbit H-Ras (MC57) (05-775 Upstate, Billerica, MA), rabbit anti-Fra1 (R-20) (sc605) and goat anti-Crumbs3 (C-15) (sc27904) (both purchased from Santa Cruz, Santa Cruz, CA), mouse anti α-tubulin (B512) (T5168 Sigma, St Louis, MO) and mouse anti-ZO1 (339100, Invitrogen, Grand Island, NY). Mouse monoclonal anti-Scribble (7C6-D10) has been previously described (Dow et al., 2003). Inhibitor experiments were performed using the MEK1 inhibitor PD98059 (20 µM) (P125 Sigma Aldrich, St Louis, MO), the PI3K inhibitor LY294002 (20 µM) (440202, Merck, Germany) and the JNK inhibitor SP600125 (20 µM) (420119 Merck).
MCF10A cells were maintained in DMEM:F12 (Dulbecco's Modified Eagle Medium: F12) supplemented with 5% donor horse serum (Gibco, NY), 10 µg/ml insulin (Novo Pharmaceuticals, UK), 0.5 µg/ml hydrocortisone (Sigma Aldrich), 20 ng/ml EGF (Cytolab Ltd, Switzerland), 100 ng/ml cholera toxin (Sigma Aldrich), penicillin (100 U/ml) and streptomycin (100 U/ml) as previously described (Debnath et al., 2003). 293T cells were maintained in DMEM supplemented with 10% foetal bovine serum, penicillin (100 U/ml) and streptomycin (100 U/ml). All cultures were maintained at 37°C in 5% CO2.
Retroviral constructs and infection
Full length human Scribble was cloned into MSCV-IRES-GFP as previously described (Dow et al., 2003). The coding sequence of Crumbs3 had previously been PCR amplified and cloned into MSCV-IRES-GFP vector using EcoRI sites. All constructs were verified by sequencing. Virus was generated by transfecting 293T cells by calcium phosphate precipitation using the amphotrophic packaging vector RD114 and appropriate DNA constructs. Stable MCF10A cell lines were generated essentially as described previously (Debnath et al., 2003). Cultured MCF10A cells were infected with virus 30, 42 and 54 hours post transfection and allowed to recover for 24 hours before selection. To generate polyclonal stable populations, cells were either selected for one week in 2 µg/ml puromycin and/or sorted for GFP+ cells on a FACStar flow cytometer (Becton Dickinson, Franklin Lakes, NJ). All stable cell lines were monitored and regularly checked for expression and re-selected if necessary.
MCF10A cells were plated out at 7.5×104 cells in 6-well dishes. 24 h later culture media was replaced with EGF-depleted media. Cells were cultured in EGF-depleted media for 48 h before replacing with pre-warmed media containing 20 ng/ml EGF. Cells were harvested for protein analysis at designated timepoints. Quantification was performed by densiometry of immunoblot bands using ImageJ software (ImageJ, NIH USA).
JNK kinase assay
MCF10A cells were grown in 10 cm dishes till they reached 80% confluency. Cells were UV irradiated (40 J/m2), washed in PBS and fresh media applied for 30 min before harvesting. JNK activity was assessed using a non-radioactive kinase assay kit (9810, Cell Signaling) according to the manufacturer's instructions.
Protein concentrations from prepared whole cell lysates were determined using a Lowry Protein Assay (BioRad, Hercules, CA) and resolved on pre-cast gradient gels (Invitrogen). Samples were transferred overnight at 4°C and probed with the relevant antibodies. Proteins were visualised using LumiLight ECL reagents (Roche, Germany) according to the manufacturer's instructions.
RNA isolation, cDNA synthesis and qRT-PCR
RNA was harvested using TRizol reagent (Invitrogen) and purified using chloroform extraction and ethanol precipitation. RNA quality and concentration was measured using a Nanodrop ND-1000 spectrophotometer (ThermoScientific, Vic, Aust). cDNA was made from 2 mg of RNA using standard methods (Dow et al., 2008). Samples were run in triplicate on a StepOnePlus Real-Time PCR System (Applied Biosystems, Carlsbad, CA). All samples were normalized to GAPDH control (or L32 when using siGAPDH) and fold change between samples was calculated using the comparative C(T) method. Primers used for qRT-PCR are listed in supplementary material Table S2. qRT-PCR experiments were performed on multiple independently derived cell lines and at least 3 independent siRNA transfections. Statistics were calculated on results from 3–6 runs using GraphPad Prism data software (GraphPad version 5.0b).
All siRNAs used were purchased from Dharmafect SMARTpool (ThermoScientific, Rockford, IL) and siGFP or siGAPDH were used as controls. Each siRNA pool contained four oligos targeting separate mRNA regions. For downstream protein or RNA applications MCF10A cells were plated out in 1.6 ml of antibiotic-free media at 1.1×105 cells per well in a 6-well dish ∼16 hours prior to transfection. Prior to transfection, siRNAs were diluted to a concentration of 50 nM in 200 µl OpitiMEM and a lipid mix made using 3 µl of Dharmafect3 lipid (ThermoScientific) and 197 µl OptiMEM. For experiments using multiple siRNAs, the highest combined concentration was used to determine the concentration of the siGFP or siGAPDH control. The lipid and siRNA mix were combined and allowed to duplex for 20 min before adding to cells. Cells were washed and fresh media applied after 24 hours. For RNA analysis, cells were harvested 48 hours following transfection and for protein analysis, cells were harvested 72 hours post transfection. For downstream immunofluorescent staining MCF10A cells were plated out in 400 µl of antibiotic-free culture media at 3.6×104 cells per well onto sterile poly-L-lysine-coated coverslips (#354085, Becton Dickinson) in a 48-well dish ∼20 hours prior to transfection. As above Dharmafect3 lipid (ThermoScientific) was used and complexed with the appropriate siRNAs at 50 nM. Fresh media was applied after 24 hours. Cells were used for subsequent immunofluorescent staining 72 hours post transfection.
Immunostaining and quantification
MCF10A cells were grown on sterile coverslips prior to fixing in 100% ice-cold methanol for 20 mins for ZO1 staining or 4% paraformaldehyde for 15 min at room temperature followed by a 10 min permeabilization step in 1% SDS for Occludin and Claudin1 staining. Coverslips were washed in PBS before blocking in 2% BSA/PBS for 1 hour at room temperature. Samples were incubated with mouse anti-ZO1 (339100, Invitrogen), rabbit anti-occludin (71-1500, Zymed) or rabbit anti-claudin1 (51-9000, Zymed) overnight at 4°C. Coverslips were washed with PBST before probing with secondary anti-mouse antibodies conjugated to Alexa 488 flurochromes (Molecular Probes, Grand Island, NY)). ProLong Gold Antifade containing DAPI (Invitrogen) was used to visualize nuclei. Images were acquired on a FV1000 BX51 scanning confocal microscope (Olympus Corp, Japan) using a 40×oil objective. To quantify tight junctions, continuous ZO1 staining was assessed using MetaMorph 6.3 software (Molecular Devices Corp., Downington, PA). Slides stained for ZO1 were viewed under a confocal microscope and multiple images taken randomly across the entire sample. Images were analyzed with MetaMorph software using the ‘Angiogenesis Tube Formation’ application. From the analysis, the value describing the ‘tube length/set’ was used as a measure of continuous ZO1 staining. This value was divided by the number of cells in the image (determined by DAPI staining). For each experiment at least 1000 cells were analyzed across multiple images. Identical analysis was carried out for controls and values expressed as fold change relative to control. Data was plotted and analyzed using GraphPad Prism data software (GraphPad version 5.0b).
Transepithelial electrical resistance measurements
Cells were seeded at 6×105 per well onto polyester transwell filters (Sigma Aldrich, CLS3450) and readings recorded every second day using a Millicell-ERS (MERS00001 Millipore, Billerica, MA) according to the manufacturer's instructions. Resistance was calculated as resistance of a unit area = resistance (Ω)×effective membrane area (cm2). Sample readings were measured in triplicate for each reading. Media was replaced following each reading.
Size-selective assessment of tight junction paracellular permeability
Cells were seeded at 6×105 or 2×105 cells per well on polyester transwell filters (Sigma Aldrich, CLS3450) for 13 or 3 days. Media was replaced every 2 days and on the day indicated TER measurements were recorded and the media was replaced with 1 ml in the bottom well and 250 µl in the top well. Fluorescently labeled dextrans were added to the top well, 25 µl of 4 kDa FitC-Dextran (Sigma, FD-4) and 25 µl of 70 kDa Rhodamine-Dextran (Sigma, R-9379) and incubated at 37°C for approx 3 hours. Media from the bottom plate was aliquoted to 3 wells in a black-walled 96-well plate and the following wavelengths; FitC (Exc: 485 nm, Em: 544 nm) and Rhodamine (Exc: 520 nm, Em: 590 nm) on a Bio-Stack automated plate reader (BioTak, Millennium Science, Australia) (Matter and Balda, 2003). The average of the readings were taken as technical replicates and the experiments were performed in biological repeats as indicated.
Transmission electron microscopy
Cells were grown to confluency before being fixed in 2% paraformaldehyde, 2.5% glutaraldehyde in 0.08 M Sorensen's phosphate buffer/PBS for 30 mins, then rinsed in PBS containing 5% sucrose. Post-fixation was carried out using 2% osmium tetroxide in PBS followed by dehydration through a graded series of alcohols, two acetone rinses and embedding in Spurrs resin. Sections ∼80 nm thick were cut with a diamond knife (Diatome, Switzerland) on an Ultracut-S ultramicrotome (Leica) and contrasted with uranyl acetate and lead citrate. Images were captured with a Megaview II cooled CCD camera (Olympus) on a JEOL 1011 transmission electron microscope.
Microarray experiments and analysis
RNA integrity was assessed using a bioanalyzer (Agilent Technologies, USA) and RNA with a RNA integrity number (RIN) of 10 were used for microarray experiments. Microarray experiments were carried out using the GeneChip Human Gene 1.0 ST Array kit (Affymetrix, Santa Clara, CA) according to the manufacturer's instructions. Quality control metrics were carried out using Expression Console software (Affymetrix) and Partek software (Partek Incorporated, USA). Gene lists were generated using Tukey's biweight 1 step summarisation and batch effect was applied to passage number and cell line. Pathway and gene ontology analysis was applied using the online DAVID bioinformatics database (http://david.abcc.ncifcrf.gov).
We thank members of our laboratory for helpful discussions. Special thanks to the Peter MacCallum Cancer Centre Microscopy Core, in particular Stephen Asquith for preparation of the TEM samples. We also thank the Peter MacCallum Cancer Centre Microarray Core in particular Dr Jason Li for help and advice with analysis.
I.A.E. designed, planned and carried out the experiments, analyzed and interpreted the data and wrote the manuscript; C.M. assisted with experimental work and approach, interpretation of data and edited the manuscript; P.O.H. designed the experiment, sourced funding, provided reagents and mentorship, edited the manuscript.
This study was supported by a project grant from the National Health and Medical Research Council of Australia [grant number APP1004434 to P.O.H.]. I.A.E was supported by a Cancer Council Victoria Postgraduate Cancer Research Scholarship and P.O.H by a Biomedical Career Development Award from the National Health and Medical Research Council of Australia.